摘要
航天器在故障定位过程中易受原始电源信号的干扰,导致识别效果较差,为了解决该问题,提出了基于多层前馈神经网络的航天器在线故障检测系统设计;根据航天器在线故障检测原理及物联网技术设计系统总体架构,并分别对硬件部分及软件部分进行设计;硬件部分结合工业标准PC组件,设计PXI机箱结构,完成对PXI测量模块的控制,利用MXI-4接口工具实现远程遥控,解决干扰信号对系统定位识别干扰;设计FPGA的EP3C10芯片外围结构,确定电路板主、子适配器管脚连接方式,利用两个高速AD转换器差分采样,通过FIFO存储采样结果;通过电子负载板继电器控制模块,控制信号阻断性能;构建基于多层前馈神经网络识别模型,依据确定性逻辑推理规则得出识别门限值,依据阈值设定具体识别流程,判断则判定参数有故障,完成系统设计;实验结果表明,该系统信号阻断效果优异,在距离为2m和6m时达到最大信号幅值0.9,故障模式的检测结果与理想结果一致,能够为航天器稳定运行提供设备支持。
In the process of fault location,the spacecraft is susceptible to the interference of the original power signal,which leads to a poor recognition effect.To solve this problem,a design of a spacecraft online fault detection system based on a multilayer feedforward neural network is proposed.According to the spacecraft online fault detection principle and the Internet of Things technology,the overall architecture of the system is designed,and the hardware and software parts are designed respectively.The hardware part combines the industry standard PC components to design the PXI chassis structure,complete the control of the PXI measurement module,and use the MXI-4 interface tool to achieve remote control to solve the interference signal to the system positioning identification interference.Design the peripheral structure of the EP3 C10 chip of FPGA,determine the connection mode of the main and sub adapter pins of the circuit board,use 2 high-speed AD converter differential sampling,and store the sampling results through the FIFO.Through the electronic load board relay control module,control signal blocking performance.A recognition model based on a multi-layer feed-forward neural network is constructed,a recognition threshold is obtained according to deterministic logic inference rules,a specific recognition process is set according to the threshold,and the judgment determines that the parameter is faulty and the system design is completed.Experimental results show that the signal blocking effect of the system is excellent,and the maximum signal amplitude is 0.9 at distances of 2 and 6 m.The detection results of the failure mode are consistent with the ideal results,which can provide equipment support for the stable operation of the spacecraft.
作者
戴峻峰
赵建
Dai Junfeng;Zhao Jian(School of Cyberspace Security,ChengduUniversity of Information Technology,Chengdu 610200,China;Electronic Engineering Practice Center,Chengdu University of Information Technology,Chengdu 610200,China)
出处
《计算机测量与控制》
2020年第8期93-97,共5页
Computer Measurement &Control
关键词
多层前馈神经网络
航天器
故障检测
物联网
multilayer feedforward neural network
spacecraft
fault detection
internet of things